Room: Stars at Night Ballroom 4
Purpose: Roughly 15% of institutions fail plan delivery tests as measured by IROC-Houstonâ€™s head-and-neck phantom. Studies have revealed that beam modeling of the multi-leaf collimator (MLC) is one important source of error, and separate studies show that these errors may not be detectable by standard QA methods. Here we investigate the impact of the treatment planning system (TPS) MLC model on dosimetric endpoints for tumor coverage and organ-at-risk (OAR) dose.
Methods: We created 9 variations of a 6MV photon beam model for an Agility MLC (Elekta Inc.) in a commercial TPS (RayStation v.6.1) with systematic shifts in the MLC leaf-tip offset parameter of up to Â±2 mm. Ten head and neck treatment plans were recalculated using these altered beam models. Each of these plans was compared to the clinical model, and the percentage dose difference was calculated as a function of leaf-tip offset. The primary endpoints for the treatment planning studies were PTV D99%, right and left parotid Dmean and spinal cord Dmax.
Results: Leaf-tip offsets of 1 mm and 2 mm resulted in PTV D99% dose differences of (6.9 Â±1.4)% and (10.0 Â± 2.1)% respectively. For the OARs, the spinal cord maximum dose differences were (7.7Â±2.1)% and (13.2Â±3.9)% for offsets of 1 mm and 2 mm respectively. The percentage mean dose differences were (10.4Â±2.1)% and (17.5Â±3.8)% for the right parotid gland and (11.0Â±2.9)% and (18.4Â±5.3)% for the left parotid gland for 1 mm and 2 mm offsets respectively.
Conclusion: The MLC model parameters in the TPS can have large impacts on the dosimetric endpoints for tumor coverage and organs-at-risk, with shifts as small as 1 mm leading to dose deviations in excess of 10% for the PTV and OARs. External validation methods may be used to identify these errors in the planning system.
Not Applicable / None Entered.